Francesco Casella

Regarding the case in question, solving linear equations may not be enough, you could need to do something like solving P/rho = RT for rho. I guess linear equations combined...

> As long as there is only scalar variables and no function calls, records, tuples, it should be fine though In our case there should be none, this is also...

@phannebohm, @linuslangenkamp can we close this ticket? Do you think #13553 is a sound implementation of the annotation, or is it just experimental?

> Yes, I guess _computationally_ the job is easy. The only thing left to do is match the inner variables with torn equations in a correct way. But that entails...

@kabdelhak, here is an MWE that reproduces the structure of the SOFC initialization model but is more manageable in terms of equations. ```modelica package TestGuruTearing package Air = Modelica.Media.IdealGases.MixtureGases.CombustionAir; model...

Let's first consider the SingleComponent model. Variables `c.w1`, `c.w5` and `c.Q` are immediately computed from time. The last three equations of `c` allow to compute all the variables of `c.oxygen`,...

The `ArraySeparate` model basically instantiates N independent component models. Hence, you should get `N` separate strong components, each with its 8 scalar (4 scalar + 2 arrays of dimension 2)...

> @casella should we backport to 4.1.1? I suppose "no". I would keep 4.1.1 for critical fixes, e.g. if we discover late that we introduced regressions in 4.1.0.

> I can understand that feature, but there's also a need to innovate and replace them with something better: > > * Instead of just storing (or not storing) protected...

> I thus think we can get agreement on setting for storing protected, and non-equidistant if we reformulate it to indicate that one wants to store transients i.e., more frequently...